The central Greek town of Orchomenos seems an unlikely place to start a debate on race and genetics in the 21st century. It is home to several archaeological relics that demonstrate the town's Bronze Age power and influence. The inhabitants of Orchomenos live with a famous tomb, a well preserved 2,400-year-old theatre and a part-excavated Mycenaean palace. They also live under the shadow of sickle cell anaemia.

Up to one in four people in Orchomenos carry the gene for the blood disorder, which is almost always associated in textbooks, and in people's minds, with black people. That's more than double the rate among African Americans.

Orchomenos is not unique: a scattering of isolated Mediterranean communities face a similarly high risk. A few thousand miles away in South Africa, the guilty gene is almost totally absent, in both black and white people.

The "black disease" label for sickle cell anaemia is not just inaccurate, it is life-threatening. In 1996, scientists reported that an eight-year-old "phenotypically-European" (white) boy with severe abdominal pain and anaemia was seen by his doctor. Medics were preparing to operate when a technician noticed strange-looking cells in the boy's blood smeared on a slide. Despite parents of Indian, northern European and Mediterranean ancestry, he had the disorder. The unnecessary operation was cancelled.

In that case, assumptions about disease on grounds of race were wrong. Yet it remains undeniably true that, on average, sickle cell anaemia is more common in black people than white people, so does it make sense to test every white person with the eight-year-old's symptoms for sickle cells? And is it wrong for doctors to assume the same symptoms in a black person are down to "their" disease - when they probably are?

Geneticists tend to avoid questions like these. Historically, biological studies of race have been tainted by eugenics. But in the geneticists' absence, others have stepped in to argue that race is irrelevant to medicine. In 2001, the New England Journal of Medicine went as far in an editorial to declare it "biologically meaningless".

Geneticists do not agree, and with information from the human genome sequence now raising the prospect of drugs tailored to particular groups, they have returned to the debate. "It's pretty hard with a bunch of scientists to say 'no you can't talk about this'," says Lynn Jorde, a geneticist at Utah University.

At the annual meeting of the American Society of Human Genetics in Toronto last month, Jorde and other researchers in the field offered their views of how their discipline should tackle the race minefield. The results are published this month as a supplement to the journal Nature Genetics.

"I react against what I think of as political correctness that says there is no such thing as race," says Kenneth Kidd, a geneticist at Yale University and another contributor. The line in the US Declaration of Independence that all men are created equal "is a statement of morality, not of science," he says. Study after study shows the genetic mutations that cause disease are not uniformly distributed around the world. "Genetic variation is not uniform, but that has nothing to do with the value of a person."

The big problem with race and genetics, Kidd says, is neither race nor genetics, but our clumsy use of language. Skin colour is a poor shortcut to race, which is not precise enough to encompass where in the world our ancestors evolved. Skin colour is affected by the environment and varies a lot within groups; what we think of as "race" is a crude attempt to group people together for social reasons that do not always tally with their geographical ancestry, and so their biological differences.

A sterotypical term like "Asian" is too broad, Kidd argues. "It might make a difference [to their genes] if the person is from Thailand or Cambodia, Japan, northern China or southern China."

The geneticists' return to the debate is well timed. This week, full results of a trial of the world's first "ethnic drug" are due to be presented at a meeting of the American Heart Association. The trial of a drug called BiDil, to treat heart failure in black people, was stopped early because the results were so encouraging. Its maker Nitromed hopes to market the drug to African Americans next year.

The increased benefit is believed to come from genetic differences in the way bodies handle nitric oxide, low blood levels of which contribute to the condition. But David Goldstein, who studies population genetics at University College London, says nobody has confirmed this, or directly compared the fate of nitric oxide inside people with their ancestry in Africa or Europe. It may be down to genetics, or subtle environmental factors may be at play.

There are also concerns that companies developing drugs for specific ethnic groups could create "orphan genotypes" - genetic groups not large or profitable enough to bother with.

So just how important is our ancestry when it comes to our medicine? Is it as meaningless as the social scientists would like, or is it critical for deciding how to treat us when we fall ill, so important that it is worth risking the social baggage it brings?

Jorde at Utah says: "People have not sat down and asked what we mean by biologically meaningless. You can't say there are no differences because I can show you that there are, so then at what level, at what point are those differences meaningless?"

An answer is still a long way off, but at least the question is back on the agenda.

· The Nature Genetics race and genes supplement is at www.nature.com/ng